Analysis of Tamoxifen and Its Metabolites by On-Line Capillary

Shahab A. Shamsi, Bertha C. Valle, Fereshteh Billiot, and Isiah M. Warner ..... Lewis L. Smith , Karen Brown , Philip Carthew , Chang-Kee Lim , Elizab...
0 downloads 0 Views 583KB Size
Download PDF
Anal. Chem. 1996, 68, 668-674

Analysis of Tamoxifen and Its Metabolites by On-Line Capillary Electrophoresis-Electrospray Ionization Mass Spectrometry Employing Nonaqueous Media Containing Surfactants Wenzhe Lu,† Grace K. Poon,‡ Paul L. Carmichael,§ and Richard B. Cole*,†

Department of Chemistry, University of New Orleans, New Orleans, Louisiana 70148, CRC Centre for Cancer Therapeutics, Institute of Cancer Research, Surrey, U.K., and Section of Molecular Carcinogenesis, Haddow Laboratories, Institute of Cancer Research, Surrey, U.K.

On-line capillary electrophoresis-electrospray ionization mass spectrometry (CE-ESMS) has been employed for the analysis of metabolites of the anticancer drug tamoxifen. Nonaqueous (methanol) CE electrolyte provided better resolution and detection sensitivity compared to aqueous systems or highly aqueous water-methanol electrolyte mixtures. Nonaqueous methanol also permitted the use of lower ES voltages presumably owing to its lower surface tension, which facilitated droplet breakup. This decreased the tendency to produce electric discharges, thus improving the stability of electrospray conditions. The relative ease of methanol solvent evaporation may contribute to an improved yield of protonated analytes as compared to highly aqueous solutions. Enhanced CE resolution can be at least partially attributed to the improved solubility of analytes in methanol relative to water. Higher solubility implies less aggregation of hydrophobic analytes, thus improving homogeneity in solution. Moreover, electroosmotic flow toward the detector decreased in methanol relative to water. The reduction of this force pushing all analytes through the capillary, but not aiding in separation, implies that other factors such as slight differences in electrophoretic mobilities are more apt to lead to successful separations. Surfactants were employed as nonaqueous CE-ESMS buffer additives. An SDS concentration of 7 mM lowered the ESMS signal response for N-desmethyltamoxifen by a factor of ∼3. However, separation of tamoxifen metabolites using 7 mM SDS was augmented relative to the unadulterated methanol electrolyte. This enabled the separation of r-hydroxytamoxifen and 4-hydroxytamoxifen, which were not resolvable in methanol electrolyte devoid of SDS. The methanol-surfactant electrolyte system has been successfully used to determine metabolites formed after incubation of tamoxifen with mouse hepatocytes. Tamoxifen [trans-1-[4-(2-dimethylamino)ethoxy]phenyl-1,2diphenylbut-1-ene] is an anti-estrogen drug that is routinely used for the treatment of breast cancer.1 Recently, it has also been employed as a preventative agent for healthy woman who have a †

University of New Orleans. CRC Centre for Cancer Therapeutics. § Haddow Laboratories. (1) Jordan, V. C. Br. J. Pharmacol. 1993, 110, 507-517. ‡

668 Analytical Chemistry, Vol. 68, No. 4, February 15, 1996

family history of breast cancer.2,3 This preventative use has been somewhat controversial because there is some evidence that administration of tamoxifen can cause liver tumors in rats.4-7 The tumors are believed to initiate upon formation of covalent tamoxifen-DNA adducts,8 and research into the effects of tamoxifen on human hepatocytes is currently under way.7 Isolation and structural characterization of metabolites produced in vivo are necessary steps to understanding the mechanism responsible for drug activity. Tamoxifen is thought to act mainly by blocking the estrogen receptor in tumor tissue, thus inhibiting tumor growth. Alternative mechanisms of action have also been suggested.9 The metabolism of tamoxifen is a complex process with many potential reaction pathways. Metabolites of the drug produced after oral ingestion have been extensively studied for female patients, rats, mice, monkeys, and dogs.10-13 On-line HPLC-MS has been used to study tamoxifen and the metabolite 4-hydroxy-N-desmethyltamoxifen.14,15 In addition, LC-MS-MS has been applied to the determination of phase I and phase II metabolites of tamoxifen in breast cancer patients.16 Compared to HPLC separation methods, capillary electrophoresis (CE) offers the potential for higher resolution and faster analysis of many pharmaceutical compounds.17 Another advantage of CE is that very small sample quantities (e.g., femtomoles) can (2) Powles, T. J.; Hardy, J. R.; Ashley, S. E.; Farrington, G. H.; Cosgrove, D.; Davey, J. R.; Dowsett, M.; McKinna, J. A.; Wash, A. G.; Sennet, H. D.; Tillyer, C. R.; Treleaven, J. G. Br. J. Cancer 1989, 60, 126-131. (3) Nayfield, S. G.; Karp, J. E.; Ford, L. G.; Dorr, F. A.; Kramer, B. S. J. Natl. Cancer Inst. 1991, 83, 1450-1459. (4) Greaves, P.; Goonetillebe, R.; Nunn, G.; Topham, J.; Orton, T. Cancer Res. 1993, 53, 3919-3924. (5) Meshter, C.; Rendall, M.; Rose, D.; Jordan, K.; Williams, G. In Proceedings of the American Society of Toxicology; 1991; Abstract 695, p 695. (6) Hirsimaki, P.; Hirsimaki, Y.; Nieminen, L.; Payne, B. J. Arch. Toxicol. 1993, 67, 49-54. (7) Potter, G. A.; McCague, R.; Jarman, M. Carcinogenesis 1994, 15, 439-442. (8) Phillips, D. H., et al. Carcinogenesis, submitted. (9) Lønning, P. E.; Lien, E. A. Cancer Surv. 1993, 17, 343-351. (10) Furr, B. J. A.; Jordan, V. C. Pharmacol. Ther. 1984, 25, 127-205. (11) Lønning, P. E.; Lien, E. A.; Lundgren, S.; Kvinnsland, S. Clin. Pharmacokinet. 1992, 22, 327-358. (12) Jordan, V. C. Pharmacol. Rev. 1984, 36, 245-276. (13) Jarman, M.; Poon, G. K.; Rowlands, M. G.; Grimshaw, R. M.; Horton, M. N.; Potter, G. A.; McCague, R. Carcinogenesis 1995, 16, 683-690. (14) Lien, E. A.; Solheim, E.; Kvinnsland, S.; Ueland, P. M. Cancer Res. 1988, 48, 2304-2308. (15) Lien, E. A.; Solheim, E.; Lea, O. A.; Lundgren, S.; Kvinnsland, S.; Ueland, P. M. Cancer Res. 1989, 49, 2175-2183. (16) Poon, G. K.; Chui, Y. C.; McCague, R.; Lønning, P. E.; Feng, R.; Rowlands, M. G.; Jarman, M. Drug Metab. Dispos. 1993, 21, 1119-1124. (17) Monning, C. A.; Kennedy, R. T. Anal. Chem. 1994, 64, 280R-413R. 0003-2700/96/0368-0668$12.00/0

© 1996 American Chemical Society

be analyzed, which is particularly important for sample-limited studies. On-line CE-MS has been recognized as a powerful method for pharmaceutical and biomedical analyses,18-21 although to our knowledge, CE-MS analysis of tamoxifen and its metabolites has not been reported. Since electrospray ionization efficiently creates gas phase ions out of nonvolatile compounds present in solution, it is considered as a preferred mass spectrometric ionization method for interfacing to CE systems. Conventional CE analyses of pharmaceutical compounds are commonly performed in aqueous buffers. Some drugs and metabolites, such as those of the antitumor drug pyrazoloacridine,19 are hydrophobic in nature and are thus less amenable to aqueous systems. Aqueous CE analyses of such compounds often encounter problems of reduced resolution and lowered throughput. These difficulties are at least partly due to the tendency of analytes to form hydrophobic aggregates. Addition of organic solvents to the CE buffer to increase the analyte solubility22-24 can help to alleviate the problem. Moreover, 100% nonaqueous solvents have been applied to CE analyses of compounds that are not readily soluble in water.19,25,26 Increasing analyte solubility via the use of nonaqueous solvents may enhance separation, resolution, and detection sensitivity. Surfactants have been used extensively as reagents in CE to enhance the separation of analytes having similar electrophoretic mobilities.27 Some tamoxifen analogs have been separated and determined by CE-UV using high concentrations of surfactants in organic solvents.28 The main problem associated with the use of surfactants for on-line CE-ESMS systems is that accumulation of surfactants can cause fouling of the ion source and suppression of analyte signals in ESMS.29 However, the severity of ion suppression caused by the presence of surfactants may be variable.29-31 This report describes the application of on-line CE-ESMS to the analysis of tamoxifen metabolites. A mixture of tamoxifen standards was initially examined to optimize CE-ESMS conditions, in particular, the composition of the nonaqueous media including surfactant concentration. The optimized method was (18) Johansson, I. M.; Pavelka, R.; Henion, J. D. J. Chromatogr. 1991, 559, 515528. (19) Tomlinson, A. J.; Benson, L. M.; Naylor, S. J. High Resolut. Chromatogr. 1994, 17, 175-186. (20) Olivares, L. A.; Nguyen, N. T.; Yonker, C. R.; Smith, R. D. Anal. Chem. 1987, 59, 1230-1232. (21) Smith, R. D.; Barinaga, C. J.; Udseth, H. R. Anal. Chem. 1988, 60, 19481952. (22) Tomlinson, A. J.; Benson, L. M.; Landers, J. P.; Scanlan, G. P.; Fang, J.; Gorrod, J. W.; Naylor, S. J. Chromatogr. 1993, 652, 417-426. (23) Naylor, S.; Benson, L. M.; Tomlinson, A. J. In CRC Handbook of Capillary ElectrophoresissA Practical Approach; Landers, J. P., Ed.; CRC Press: Boca Raton, FL, 1993; Chapter 17. (24) Salomon, K.; Burg, D. S.; Helmer, J. C. J. Chromatogr. 1991, 549, 375385. (25) Sahota, R. S.; Khaledi, M. G. Anal. Chem. 1994, 66, 1141-1146. (26) Walbroehl, Y.; Jorgenson, J. W. Anal. Chem. 1986, 58, 479-481. (27) Terabe, S.; Otsuka, K.; Ichikawa, K.; Tsuchiya, A.; Ando, T. Anal. Chem. 1984, 56, 111-113. (28) Ng, C. L.; Lee, H. K.; Li, S. F. Y. J. Liq. Chromatogr. 1994, 17, 3847-3857. (29) Varghese, J.; Cole, R. B. J. Chromatogr. A 1993, 652, 369-376. (30) Kirby, D.; Greve, K. F.; Flet, F.; Vouros, P.; Karger, B. L.; Nashabeh, W. Capillary Electrophoresis-Electrospray Ionization Mass Spectrometry Utilizing Background Electrolytes Containing Surfactants. Proceedings of the 42nd Conference on Mass Spectrometry and Allied Topics, Chicago, IL, May 29June 3, 1994. (31) Wendsjo, S.; Ornskov, E. Presented at the 6th International Symposium on High Performance Capillary Electrophoresis, San Diego, CA, January 31February 3, 1994.

then applied to the analysis of tamoxifen and its metabolites formed after incubation of tamoxifen with mouse hepatocytes. EXPERIMENTAL SECTION Chemicals. Tamoxifen and its analogs were synthesized according to previously reported procedures.32-34 HPLC-grade water was purchased from EM Science (Gibbstown, NJ). Methanolbased CE media were prepared by dissolving ammonium acetate (5 mM) and acetic acid (100 mM) in HPLC-grade methanol (EM Science). The surfactant-containing media were prepared by adding sodium dodecyl sulfate (SDS; Fluka, Buchs, Switzerland), Mega 10, or Genapol+C-100 from Calbiochem (La Jolla, CA) to 2.5 mM ammonium acetate-50 mM acetic acid in methanol. Preparation and Treatment of Rat Microsomes and Mouse Hepatocytes. Biological samples were prepared at the Institute of Cancer Research, Surrey, U.K. Sample availability constraints resulted in the analyses of two sample types from separate biological sources. The method of preparing rat liver microsomes was based on a procedure described previously.32 Rat microsomes (1 mL, equivalent to 15-20 mg of protein), 2.5 mM tamoxifen, 0.1 M phosphate buffer (3 mL, pH 7.4), NADPH regenerating system (1 mL), and substrate (0.5 mg) were incubated at 37 °C for 90 min. At the end of the incubation, NaCl (1 g) was added to the mixture and the pH was adjusted to 9.0 with 1 M NaOH. Substrate and metabolites were subsequently extracted with ethyl acetate. Mouse hepatocytes were isolated from the livers of female mice according to standard procedures.35 Cells were established in culture and treated with tamoxifen (0.05 mM) for 18 h, afterwhich, cells were harvested. Media from hepatocyte incubations (1 mL) were extracted with 2% ethanol in hexane (2 × 1 mL). The organic fractions were combined and concentrated to dryness. Prior to CE injection, the dry rat liver microsome sample was dissolved in 2 mL of methanol, while the dry mouse hepatocyte sample was dissolved in 0.4 mL of methanol. No sample filtering was performed prior to CE-ESMS analysis. Capillary Electrophoresis. CE separations were performed using a CE system I (Dionex Corp., Sunnyvale, CA). Fused-silica capillaries (100 µm i.d. × 245 µm o.d.) obtained from Polymicro Technologies Inc. (Phoenix, AZ) were employed in all experiments. Before use, the new capillaries were washed sequentially with water for 10 min, 0.1 M NaOH for 10 min, and water again for 20 min, followed by operating electrolyte for 20 min. For the methanol and the methanol-surfactant electrolyte systems, the capillary was washed with water for 30 min, and then operating electrolyte for 10 min, between successive runs. The electrolyte in the CE source vial was replaced every other run to minimize chemical contamination and concentration changes due to evaporation. When not in use, the capillaries were stored by placing both ends of the capillary in water with a height differential to maintain electrolyte flow through the capillary. Electrospray Mass Spectrometry. A Vestec 201 electrospray mass spectrometer (PerSeptive Biosystems, Vestec, Houston, TX) was employed for on-line CE-ESMS experiments. Collision(32) McCague, R.; Seago, A. Biochem. Pharmacol. 1986, 35, 827-834. (33) Foster, A. B.; Griggs, L. J.; Jarman, M.; van Maanen, J. M. S.; Schulten, H.-R. Biochem. Pharmacol. 1980, 29, 2077-2079. (34) Foster, A. B.; Jarman, M.; Leung, O.-T.; McCague, R.; Leclercq, G. Devleeschouwer, N. J. Med. Chem. 1985, 28, 1491-1497. (35) Berry, M. N.; Edwards, A. M.; Barritt, G. J. Isolated hepatocytes. Preparation, properties and applications. Laboratory techniques in biochemistry and molecular biology; Elsevier: Amsterdam, 1991; Vol. 21.

Analytical Chemistry, Vol. 68, No. 4, February 15, 1996

669

Figure 1. Structures of tamoxifen and metabolites.

induced dissociations were minimized in all experiments by maintaining a low skimmer-collimator voltage difference (7 V). The full-scan mode was employed in all experiments, using a scan rate of 3 s/scan, over a scan range of m/z 100-600. A Sage syringe pump (Orion Research, Boston, MA) was used for the delivery of the sheath flow of liquid during CE-ESMS. CE-ESMS Interface. The CE-ESMS interface used in this study has been described elsewhere.36 The coaxial sheath flow21 solution was 1% acetic acid in methanol, delivered by a 1 mL syringe at a rate of 2.5 µL/min through the 350 µm i.d., 27.5 cm length, stainless-steel needle surrounding the 245 µm o.d. CE capillary. The CE capillary tip was positioned ∼0.1 mm outside the stainless-steel needle to maximize sensitivity.36 Prior to each CE-ESMS run, MS operating conditions were optimized by adjusting the needle-counterelectrode distance, chamber temperatures, liquid sheath flow rates, and applied electrospray potentials while the signal of an analyte species that was infusing directly through the CE capillary into the ES source was monitored. The employed voltages at the electrospray needle and flat plate counterelectrode were 2.48 kV and 300 V, respectively. CE-ESMS runs were performed under these same optimized conditions. RESULTS AND DISCUSSION Aqueous electrolytes were initially employed for the CEESMS determination of a mixture of tamoxifen and authentic analogs. In these studies, the volatile buffer constituents ammonium acetate37-39 or acetic acid40,41 were used to preserve ESMS (36) Varghese, J.; Cole, R. B. J. Chromatogr. 1993, 639, 303-316.

670

Analytical Chemistry, Vol. 68, No. 4, February 15, 1996

sensitivity and minimize ion source fouling problems which are created when nonvolatile electrolytes are used. Figure 1 shows the structures of the compounds used in this study: tamoxifen, 4-hydroxy-N-desmethyltamoxifen, N-desmethyltamoxifen, tamoxifen N-oxide, tamoxifen epoxide, R-hydroxytamoxifen, 4-hydroxytamoxifen, and 4-hydroxytamoxifen N-oxide. The first CE electrolyte tested in this study was 10 mM ammonium acetate (aqueous). Protonated molecules (MH+) were detected by the mass spectrometer, but poor CE resolution and poor reproducibility of migration times were observed. When 30% acetic acid (aqueous) electrolyte (pH 2.0) was used, slightly better sensitivities and separation efficiencies were obtained. Still, only three separated peaks were observed from an injection of the mixture of seven standards due to overlapping migration profiles. Sensitivity was improved presumably by increased protonation of the analytes in the lower pH electrolyte.42 Improved peak separation was probably due to the increased dependence of migration times on (differing) electrophoretic mobilities of individual species. In low-pH solutions, the net positive charge associated with analyte molecules is raised, and the electroosmotic flow contribution to the overall mobility is lowered.41 (37) Deterding, L. J.; Parker, C. E.; Perkins, J. P.; Moseley, M. A.; Jorgenson, J. W. J. Chromatogr. 1991, 544, 329-338. (38) Moseley, M. A.; Jorgenson, J. W.; Shabanowitz, J.; Hunt, D. F.; Tomer, K. B. J. Am. Soc. Mass Spectrom. 1992, 3, 289-297. (39) Cole, R. B.; Varghese, J.; McCormick, R. M.; Kadlecek, D. J. Chromatogr. A 1994, 680, 363-373. (40) Thibault, P.; Paris, C.; Pleasance, S. Rapid Commun. Mass Spectrom. 1991, 5, 484-491. (41) Lu, W.; Yang, G.; Cole, R. B. Electrophoresis 1995, 16, 487-492. (42) Wang, G; Cole, R. B. Org. Mass Spectrom. 1994, 29, 417-427.

On-Line CE-ESMS in Aqueous Methanol and Nonaqueous Methanol Electrolytes. A mixed organic-aqueous solvent consisting of 20%:50%:30% methanol-water-acetic acid (v:v:v) (i.e., 5.0 M acetic acid) was tested as operating electrolyte for on-line CE-ESMS analysis of the same compounds. Results from this medium revealed higher sensitivity, more stable electrospray conditions, and better CE separation (five of seven components were resolved) than the previous aqueous system. It is known that higher stability electrospray ionization conditions are obtained from methanol-water mixtures than from pure aqueous systems because the lower surface tension permits droplet breakup at lower applied voltages, thus decreasing the tendency to produce unwanted electric discharges.43 The relative ease of solvent evaporation in the methanol-water mixtures may result in an improved desorption efficiency of protonated analytes. Enhanced resolution can be at least partially attributed to the higher solubility of analytes in the mixed solvent. Higher solubility means less aggregation of hydrophobic analytes and a more uniform distribution in solution. As a consequence, the analytes will interact more efficiently with solvent molecules and supporting electrolytes.20,21 Addition of organic solvent will also decrease the electroosmotic flow rate. A reduced electroosmotic flow toward the detector can improve resolution by allowing the factors that lead to different analyte migration velocities to work longer. The decrease of the electroosmotic flow is due in large part to the decrease in ζ potential (weak double layer) in organic solvents having low dielectric constants relative to water. Tamoxifen and its analogs have higher solubilities in organic solvents as compared to water or low organic content waterorganic mixtures. In an attempt to further improve the tamoxifen analysis, the same mixture was assayed in a methanol solution devoid of water. When CE electrolyte properties change, of course, both the electroosmotic flow and the electrophoretic mobility of analytes are affected. A standard mixture of tamoxifen and its analogs was run by on-line CE-ESMS using methanol containing 5 mM ammonium acetate and 0.6% (100 mM) glacial acetic acid as the nonaqueous CE media. Figure 2 shows the results obtained from a single run of the mixture of seven standards. Detection sensitivity was significantly increased compared to aqueous media; i.e., the signal-to-noise ratios for protonated tamoxifen metabolites in methanol were ∼5 times higher than those of equal concentrations in the aqueous (20% methanol with 30% glacial acetic acid) mixture. This improvement is presumably due to a higher ionization efficiency attributable to a lower surface tension of the electrospray droplet, and more rapid solvent evaporation, compared to aqueous mixtures. Methanol also offers much lower viscosity, i.e., viscosity is ∼0.54 cP for pure methanol, while it is ∼1.6 cP for 20% methanol in water. This is likely to be the origin of an increase in the electrophoretic mobility, as a decreased viscosity would accelerate zone migration. It also should be noted that methanol containing 1% acetic acid was used as the sheath liquid at the CE-ESMS interface for all experiments. As a result, methanol content was not diminished after sheath liquid mixing at the electrospray needle tip. In nonaqueous methanol electrolyte, the last tamoxifen analog had completely exited the capillary after ∼36.1 min, which was comparable to results obtained employing highly aqueous watermethanol electrolyte mixtures (35.7 min). The separation ef(43) Ikonomou, M. G.; Blades, A. T.; Kebarle, P. J. Am. Soc. Mass Spectrom. 1991, 2, 497-505.

Figure 2. Selected-ion electropherograms showing raw data from on-line CE-ESMS analysis of tamoxifen and six analog standards: (a) tamoxifen (MH+, m/z 372), (b) 4-hydroxy-N-desmethyltamoxifen (MH+, m/z 374), (c) N-desmethyltamoxifen (MH+, m/z 358), (d) isomers tamoxifen epoxide, 4-hydroxytamoxifen, and tamoxifen Noxide (in migration order, MH+, m/z 388), and (e) 4-hydroxytamoxifen N-oxide (MH+, m/z 404). Experimental conditions: sample concentration, each component 100 ng/µL; CE voltage, 25 kV; capillary, 100 µm i.d. × 100 cm; electrolyte, 5 mM ammonium acetate and 100 mM acetic acid in methanol; gravity injection, 15 cm for 2 s; ES voltage, 2.48 kV; MS scan rate, 3 s/scan.

ficiency was greatly improved in nonaqueous media, although baseline resolution was not achieved for all components; e.g., 4-hydroxy-N-desmethyltamoxifen (Figure 2b, 23.35 min) and 4-hydroxytamoxifen (Figure 2d, 23.65 min) still gave overlapping peaks. Thus, concurrent with the reduction in electroosmotic flow, other factors acted to increase electrophoretic mobilities and enhance separation. The increased analyte solubility in the methanolic solution may reduce deleterious effects caused by selfaggregation of analyte species, while also diminishing band broadening and sample loss due to capillary wall adsorption. An extract sample obtained after incubation of tamoxifen with rat microsomes and a control sample (blank) containing no tamoxifen were analyzed by on-line CE-ESMS, again using 5 mM ammonium acetate and 0.6% (100 mM) glacial acetic acid in nonaqueous methanol as the CE electrolyte. The CE-ESMS selected-ion electrophoregram of rat microsomal incubations is shown in Figure 3, employing conditions identical to those used in obtaining Figure 2. The detected metabolites (observed as Analytical Chemistry, Vol. 68, No. 4, February 15, 1996

671

Figure 4. Dependence of ESMS abundances of protonated tamoxifen metabolites (MH+) on the concentration of SDS in methanol media. Abundance values (arbitrary units) were obtained by averaging mass spectral signals for 3 min. Concentrations of tamoxifen metabolites contained in the standard mixture were as follows: solid circles, tamoxifen 30 ng/µL; hollow squares, N-desmethyltamoxifen 20 ng/µL; and solid triangles, 4-hydroxytamoxifen N-oxide 10 ng/µL.

Figure 3. Selected-ion electropherograms from on-line CE-ESMS analysis of an extract obtained after incubation of tamoxifen with rat microsomes: (a) tamoxifen (MH+, m/z 372), (b) 4-hydroxy-N-desmethyltamoxifen (MH+, m/z 374), (c) N-desmethyltamoxifen (MH+, m/z 358), (d) 4-hydroxytamoxifen (MH+, m/z 388, shorter migration time) and tamoxifen N-oxide (MH+, m/z 388, longer migration time), and (e) 4-hydroxytamoxifen N-oxide (MH+, m/z 404). Experimental conditions are the same as for Figure 2.

protonated molecules) include N-desmethyltamoxifen, 4-hydroxytamoxifen, tamoxifen N-oxide, and 4-hydroxytamoxifen N-oxide. Compared to the separation of the standards (Figure 2), slightly longer migration times (approximately 2-10%) were observed for these compounds. One factor which may contribute to longer migration times is that additional components contained in the biological matrix may increase the solution viscosity. These biological compounds are also likely to contribute to peak broadening relative to the electropherogram of the pure standards (Figure 2) as observed in the most pronounced fashion for protonated tamoxifen (Figure 3a). The biological matrix is suspected to be the principal factor responsible for peak broadening rather than capillary overload because similar peak shapes were observed when smaller amounts of sample were injected. It should be noted that similarities in peak shapes between m/z 374 (Figure 3b) and 372 (Figure 3a), suggest that the combination of the tamoxifen [MH+ + 2] isotope peak (m/z 374) and a poorly resolved [MH+ + 1] isotope peak (m/z 373) are likely to be contributing considerably to the observed m/z 374 signal, normally indicative of protonated 4-hydroxy-N-desmethyltamoxifen. Although these compounds were readily separated in Figure 2b (where a small tamoxifen isotope contribution does appear at 22.05 min), the broadened protonated tamoxifen peak in Figure 3 does not allow accurate assessment of 4-hydroxy-N-desmethyltamoxifen in the rat microsomes sample. The control sample (not shown) was devoid of peaks. The selected-ion electropherogram of m/z 388 (Figure 3d) shows the detection of only two protonated components whose 672 Analytical Chemistry, Vol. 68, No. 4, February 15, 1996

migration times correspond to those of 4-hydroxytamoxifen (25.4 min) and tamoxifen N-oxide (38.1 min). In a separate CE-ESMS run using the same operating conditions, R-hydroxytamoxifen (MH+ also m/z 388) and 4-hydroxytamoxifen standards were not resolvable (result not shown). However, a previous LC-MSMS study indicated that R-hydroxytamoxifen and 4-hydroxytamoxifen were both present in rat microsomal incubations.13 As a result, in addition to tamoxifen N-oxide, it is likely that R-hydroxytamoxifen and 4-hydroxytamoxifen exist in this sample as comigrating components exiting at 25.4 min (Figure 3d). On-Line CE-ESMS Using Methanol-Surfactant Electrolyte. Because the use of aqueous methanol and nonaqueous methanol electrolytes could not separate R-hydroxytamoxifen and 4-hydroxytamoxifen, whose protonated forms are isomeric at m/z 388, further improvements in the CE separation were required. Many kinds of surfactants have been used to enable separation of otherwise poorly resolved analytes in CE.44 Reports of the use of surfactants in on-line CE-ESMS date back to 1993,29 although many surfactants which commonly aid CE-UV separations have deleterious effects on CE-ESMS detection. Problems arise because of reduced sensitivity and ion source fouling. Certain surfactants, such as Genapol+C-100 and Mega-10, have been reported to exert little analyte ion suppression in ESMS, while offering much improved CE separations.30 Mega-10 and Genapol+C-100 were tested independently as surfactant additives for the nonaqueous methanol electrolyte in CE-ESMS determinations of tamoxifen analogs. Under a range of surfactant concentrations (i.e., Mega-10 or Genapol+C-100 from 2 to 40 mM), R-hydroxytamoxifen and 4-hydroxytamoxifen were still not separated, even though no significant analyte ion suppression was observed. Although more widely used in CE separations, SDS surfactant can cause severe ion suppression for some analyte ions, so it is generally not suitable for CE-ESMS. To evaluate the ion suppression effect of SDS, direct-infusion ESMS of tamoxifen standards was carried out. Figure 4 shows the ESMS signal dependence of a mixture of tamoxifen metabolite (44) Terabe, S. J. Pharm. Biomed. Anal. 1992, 10, 705-715.

Figure 5. Selected-ion electropherograms showing raw data from on-line CE-ESMS analysis of tamoxifen and four analog standards: (a) total ion current (m/z 100-600), (b) tamoxifen (MH+, m/z 372), (c) isomers R-hydroxytamoxifen and 4-hydroxytamoxifen, and tamoxifen N-oxide (in migration order, MH+, m/z 388), and (d) 4-hydroxytamoxifen N-oxide (MH+, m/z 404). Experimental conditions: sample concentration, each component 300 ng/µL; CE voltage, 25 kV; capillary, 100 µm i.d. × 100 cm; electrolyte, 7 mM SDS, 2.5 mM ammonium acetate, and 50 mM acetic acid in methanol; gravity injection, 15 cm for 2 s; ES voltage, 2.48 kV; MS scan, m/z 100-600 at 3 s/scan.

standards on SDS concentration in the nonaqueous methanol electrolyte. It can be seen that the abundances of these protonated tamoxifen analogs decreased steadily when SDS concentration changed from 0 to 10 mM, but leveled off between 10 and 30 mM. Notably, even at the highest SDS concentration, no (M + Na)+ peaks appeared for any of the tamoxifen analogs. In these experiments, the N-desmethyltamoxifen concentration was ∼20 ng/µL. When the SDS concentration was 7 mM, the peak intensity of protonated N-desmethyltamoxifen was about one-third of that observed in the absence of SDS. For a detection limit defined as S/N ) 3, the lower limit of detection of N-desmethyltamoxifen in 7 mM SDS would be 10 pg/µL or 28 pmol/µL. In a literature report,30 SDS suppression of ESMS signals from myoglobin and IGF-I was also observed, e.g., complete suppression of 6 pmol/µL myoglobin and 1.4 pmol/µL IGF-I at an SDS concentration of 6 mM. The fact that positive ion ESMS was used for detection of tamoxifen metabolites likely reduced the suppression activity of SDS relative to the negative-ion mode. SDS, of course, is an anionic surfactant that exhibits a rather weak ability to capture protons. Notably, the electrospray source provided satisfactory operation for at least one full day with constant use of the surfactant-containing electrolyte before cleaning was necessary. The surfactant-containing nonaqueous electrolyte was used in an attempt to improve the separation of isomeric R-hydroxyta-

Figure 6. Selected-ion electropherograms from on-line CE-ESMS analysis of an extract obtained after incubation of tamoxifen with mouse hepatocytes: (a) tamoxifen (MH+, m/z 372), (b) 4-hydroxyN-desmethyltamoxifen (MH+, m/z 374), (c) N-desmethyltamoxifen (MH+, m/z 358), (d) an unknown, R-hydroxytamoxifen, and 4-hydroxytamoxifen (in migration order, MH+, m/z 388), and (e) 4-hydroxytamoxifen N-oxide (MH+, m/z 404, not detected). Experimental conditions are the same as listed in Figure 5.

moxifen and 4-hydroxytamoxifen peaks, which had comigrated in the absence of surfactant additive. These two compounds and a third isomer, tamoxifen N-oxide (MH+ also m/z 388), were tested in a standard mixture also containing tamoxifen and 4-hydroxytamoxifen N-oxide. Figure 5 shows results of the CEESMS determination in electrolyte comprised of 7 mM SDS, 0.3% (50 mM) glacial acetic acid, and 2.5 mM ammonium acetate in nonaqueous methanol. The R-hydroxytamoxifen and 4-hydroxytamoxifen peaks were indeed resolvable (Figure 5c), while the ability to separate other tamoxifen analogs was also preserved, as verified in Figure 5b and d, and in followup experiments. In order to compensate for the increase in ionic strength resulting from addition of SDS, lower concentrations of glacial acetic acid and ammonium acetate were used. High solution ionic strengths lead to large CE currents (e.g., >18 µA) that cause unstable electrospray conditions.41 The critical micelle concentration (cmc) for SDS in aqueous media is 8 mM. If micelles can be formed in the nonaqueous methanol solution, the cmc will be much higher than in aqueous media because methanol is less polar than water. In polar solvents, the hydrophobic tail groups tend to aggregate, Analytical Chemistry, Vol. 68, No. 4, February 15, 1996

673

forming a hydrophobic center, while the charged head groups extend outward toward the solvent. As a result, hydrophobic solutes interact more strongly with the interior of the micelle, which is acting as a pseudostationary phase.27,44 Because the SDS is an anionic surfactant, any formed micelles exhibit electrophoretic mobility toward the positive voltage end (source vial). Hydrophobic solutes interacting with the interior of the micelle thus migrate more slowly than otherwise comparable hydrophilic compounds. Our results indicate that even when the surfactant concentration in the operating buffer is low enough such that micelles are not present, surfactant monomers still interact with analytes to enhance electrophoretic mobility differences and improve analyte separation. The employed condition appears to offer an acceptable balance between improvements in CE separation and analyte signal suppression in ESMS. This surfactant-methanol electrolyte [7 mM SDS, 0.3% (50 mM) glacial acetic acid, and 2.5 mM ammonium acetate in methanol] was used for the on-line CE-ESMS analysis of an extract of the incubation of tamoxifen with mouse hepatocytes, along with a control (blank) sample (not shown). Figure 6 displays the selected-ion electropherogram obtained from the extract under the same CE-ESMS conditions as used to obtain Figure 5. The CE-ESMS selected-ion electropherogram for m/z 388 (Figure 6d) shows well-separated R-hydroxytamoxifen (38.6 min) and 4-hydroxytamoxifen (39.25 min) peaks. An unknown component at m/z 388 with a shorter retention time (36.65 min) was also detected (Figure 6d). In addition to the above isomers, N-desmethyltamoxifen (m/z 358, Figure 6c) and residual protonated parent tamoxifen (m/z 372, Figure 6a) were also detected in the incubation sample extract from mouse hepatocytes. The small peak appearing just after the protonated tamoxifen peak in Figure 6a may arise from an unknown component in the biological matrix. The presence of 4-hydroxy-N-desmethyltamoxifen (MH+, m/z 374, Figure 6b) is less definite, owing to the fact that isotopic peaks from tamoxifen (i.e., [MH+ + 2] and a poorly resolved

674

Analytical Chemistry, Vol. 68, No. 4, February 15, 1996

[MH+ + 1]) may be contributing to the signal observed at m/z 374 (in a manner analogous to the previous discussion of Figure 3b). CONCLUSION Nonaqueous methanolic electrolyte solutions containing the surfactant SDS have proven to be useful for the CE-ESMS analysis of tamoxifen and its metabolites. Solubility appears to play a key role in the improvement of CE separations achieved using organic solvents. Higher solubility provides more homogeneous solutions by decreasing self-aggregation of hydrophobic analytes and reducing the propensity for capillary wall adsorption. Electroosmotic flow is also reduced in nonaqueous methanol relative to water. Migration times thus depend more directly on electrophoretic mobilities which have increased due to the lower solution viscosity. Furthermore, use of methanol as the electrolyte solvent also improves ESMS detection sensitivity, likely due to higher analyte throughput and increased solvent evaporation leading to a higher ionization efficiency. The CE resolution of tamoxifen metabolites has been further improved by adding SDS to methanolic electrolyte, although the presence of SDS tends to lower ESMS detection sensitivity. However, a compromise can be reached to offer acceptable detection ability and adequate separation from a single CE-ESMS condition. ACKNOWLEDGMENT The authors thank Dionex Corp. for providing CE instrumentation for this project, and A. Hewer and K. Cole for assistance in the preparation and culturing of mouse hepatocytes. Received for review August 3, 1995. Accepted November 30, 1995.X AC950786X X

Abstract published in Advance ACS Abstracts, January 15, 1996.